on a problem of halmos: unitary equivalence of a matrix to ... · on a problem of halmos: unitary...
TRANSCRIPT
On a problem of Halmos:Unitary equivalence of a matrix to its transpose
Stephan Ramon Garcia
Pomona CollegeClaremont, California
March 18, 2011
Abstract
Halmos asked whether every square complex matrix is unitarily equivalent to its transpose (UET).Ad hoc examples indicate that the answer is no. In this talk, we give a complete characterizationof matrices which are UET. Surprisingly, the naıve conjecture that a matrix is UET if and only if itis unitarily equivalent to a complex symmetric (i.e., self-transpose) matrix is true in dimensionsn ≤ 7 but false for n ≥ 8. In particular, unexpected building blocks begin to appear in dimensions6 and 8. This is joint work with James E. Tener (UC Berkeley).
Partially supported by NSF Grant DMS-1001614 (SRG) and a NSF Graduate Research Fellowship (JET).
[email protected] http://pages.pomona.edu/~sg064747
A “Halmos Problem”
Problem (Halmos, LAPB - Problem 159)
Is every square matrixunitarily equivalent to itstranspose (UET)?
Solution (Halmos)
Ad-hoc methods show that0 1 00 0 20 0 0
is not UET. Can we characterize those matrices which are UET?
[email protected] http://pages.pomona.edu/~sg064747
A “Halmos Problem”
Problem (Halmos, LAPB - Problem 159)
Is every square matrixunitarily equivalent to itstranspose (UET)?
Solution (Halmos)
Ad-hoc methods show that0 1 00 0 20 0 0
is not UET.
Can we characterize those matrices which are UET?
[email protected] http://pages.pomona.edu/~sg064747
A “Halmos Problem”
Problem (Halmos, LAPB - Problem 159)
Is every square matrixunitarily equivalent to itstranspose (UET)?
Solution (Halmos)
Ad-hoc methods show that0 1 00 0 20 0 0
is not UET. Can we characterize those matrices which are UET?
[email protected] http://pages.pomona.edu/~sg064747
Who Cares?
Why is this interesting?
It’s technically a “Halmos Problem”
Every square matrix is similar to its transpose, making theproblem somewhat difficult.
Linear preservers of the numerical range are of the formX 7→ UXU∗ or X 7→ UX tU∗ (where U is unitary). Thus wecharacterize the fixed points of linear preservers in thenontrivial case.
It has an obvious “answer,” which is totally wrong.
Funny things happen once you hit dimensions 6 and 8.
[email protected] http://pages.pomona.edu/~sg064747
Who Cares?
Why is this interesting?
It’s technically a “Halmos Problem”
Every square matrix is similar to its transpose, making theproblem somewhat difficult.
Linear preservers of the numerical range are of the formX 7→ UXU∗ or X 7→ UX tU∗ (where U is unitary). Thus wecharacterize the fixed points of linear preservers in thenontrivial case.
It has an obvious “answer,” which is totally wrong.
Funny things happen once you hit dimensions 6 and 8.
[email protected] http://pages.pomona.edu/~sg064747
Who Cares?
Why is this interesting?
It’s technically a “Halmos Problem”
Every square matrix is similar to its transpose, making theproblem somewhat difficult.
Linear preservers of the numerical range are of the formX 7→ UXU∗ or X 7→ UX tU∗ (where U is unitary). Thus wecharacterize the fixed points of linear preservers in thenontrivial case.
It has an obvious “answer,” which is totally wrong.
Funny things happen once you hit dimensions 6 and 8.
[email protected] http://pages.pomona.edu/~sg064747
Who Cares?
Why is this interesting?
It’s technically a “Halmos Problem”
Every square matrix is similar to its transpose, making theproblem somewhat difficult.
Linear preservers of the numerical range are of the formX 7→ UXU∗ or X 7→ UX tU∗ (where U is unitary). Thus wecharacterize the fixed points of linear preservers in thenontrivial case.
It has an obvious “answer,” which is totally wrong.
Funny things happen once you hit dimensions 6 and 8.
[email protected] http://pages.pomona.edu/~sg064747
Who Cares?
Why is this interesting?
It’s technically a “Halmos Problem”
Every square matrix is similar to its transpose, making theproblem somewhat difficult.
Linear preservers of the numerical range are of the formX 7→ UXU∗ or X 7→ UX tU∗ (where U is unitary). Thus wecharacterize the fixed points of linear preservers in thenontrivial case.
It has an obvious “answer,” which is totally wrong.
Funny things happen once you hit dimensions 6 and 8.
[email protected] http://pages.pomona.edu/~sg064747
Who Cares?
Why is this interesting?
It’s technically a “Halmos Problem”
Every square matrix is similar to its transpose, making theproblem somewhat difficult.
Linear preservers of the numerical range are of the formX 7→ UXU∗ or X 7→ UX tU∗ (where U is unitary). Thus wecharacterize the fixed points of linear preservers in thenontrivial case.
It has an obvious “answer,” which is totally wrong.
Funny things happen once you hit dimensions 6 and 8.
[email protected] http://pages.pomona.edu/~sg064747
A naive conjecture
Definition
If T is unitarily equivalent to a complex symmetric (i.e., self-transpose) matrix, then T is UECSM.
A Coincidence?
Halmos’ matrix 0 1 00 0 20 0 0
is the “simplest” example of a matrix which is not UECSM.
Naive Conjecture
UET ⇐⇒ UECSM (⇐ trivial)
[email protected] http://pages.pomona.edu/~sg064747
A naive conjecture
Definition
If T is unitarily equivalent to a complex symmetric (i.e., self-transpose) matrix, then T is UECSM.
A Coincidence?
Halmos’ matrix 0 1 00 0 20 0 0
is the “simplest” example of a matrix which is not UECSM.
Naive Conjecture
UET ⇐⇒ UECSM (⇐ trivial)
[email protected] http://pages.pomona.edu/~sg064747
A naive conjecture
Definition
If T is unitarily equivalent to a complex symmetric (i.e., self-transpose) matrix, then T is UECSM.
A Coincidence?
Halmos’ matrix 0 1 00 0 20 0 0
is the “simplest” example of a matrix which is not UECSM.
Naive Conjecture
UET ⇐⇒ UECSM (⇐ trivial)
[email protected] http://pages.pomona.edu/~sg064747
UECSMs are Tricky
Exactly one of the following is UECSM:
(0 7 00 1 20 0 6
) (0 7 00 1 30 0 6
) (0 7 00 1 40 0 6
) (0 7 00 1 50 0 6
) (0 7 00 1 60 0 6
)
Which one is UECSM?
The fourth matrix is unitarily equivalent to2 +
√572
0 − 12i
√37− 73
√2
57
0 2−√
572
− 12i
√37 + 73
√2
57
− 12i
√37− 73
√2
57− 1
2i
√37 + 73
√2
573
[email protected] http://pages.pomona.edu/~sg064747
UECSMs are Tricky
Exactly one of the following is UECSM:
(0 7 00 1 20 0 6
) (0 7 00 1 30 0 6
) (0 7 00 1 40 0 6
) (0 7 00 1 50 0 6
) (0 7 00 1 60 0 6
)
Which one is UECSM?
The fourth matrix is unitarily equivalent to2 +
√572
0 − 12i
√37− 73
√2
57
0 2−√
572
− 12i
√37 + 73
√2
57
− 12i
√37− 73
√2
57− 1
2i
√37 + 73
√2
573
[email protected] http://pages.pomona.edu/~sg064747
UECSMs are Tricky
Exactly one of the following is UECSM:
(0 7 00 1 20 0 6
) (0 7 00 1 30 0 6
) (0 7 00 1 40 0 6
) (0 7 00 1 50 0 6
) (0 7 00 1 60 0 6
)
Which one is UECSM?
The fourth matrix is unitarily equivalent to2 +
√572
0 − 12i
√37− 73
√2
57
0 2−√
572
− 12i
√37 + 73
√2
57
− 12i
√37− 73
√2
57− 1
2i
√37 + 73
√2
573
[email protected] http://pages.pomona.edu/~sg064747
A Gallery of UECSMs
3 6 60 4 10 0 3
∼= 1
2
(7−√
74)
3i
√2
109
(74−
√74)
12i
√1
109
(2629 + 72
√74)
3i
√2
109
(74−
√74)
3 + 36√
74109
3109
√5476 + 218
√74
12i
√1
109
(2629 + 72
√74)
3109
√5476 + 218
√74 1
218
(763 + 37
√74)
3 5 4
0 8 40 0 3
∼= 1
2
(11−
√82)
2i
√2
73
(82− 5
√82)
12i
√1
73
(1537 + 160
√82)
2i
√2
73
(82− 5
√82)
3 + 16√
8273
273
√6724 + 730
√82
12i
√1
73
(1537 + 160
√82)
273
√6724 + 730
√82 1
146
(803 + 41
√82)
5 2 2
7 0 07 0 0
∼= 1
2
(5 −√
187)
−5i
√561+5
√187
1658−i
√3350829− 125
√187
1658
−5i
√561+5
√187
16581
829
(1870 + 293
√187)
9829
√12
(173723 + 7075
√187)
−i
√3350829− 125
√187
16589
829
√12
(173723 + 7075
√187)
81−5+3
√187
9 8 9
0 7 00 0 7
∼= 8−
√1492
92 i
√16837+64
√149
13093 i
√13367213093 −
1296√
14913093
92 i
√16837+64
√149
13093207440+9477
√149
2618618√
3978002+82324√
14913093
i
√13367213093 −
1296√
14913093
18√
3978002+82324√
14913093
92675+1808√
14913093
[email protected] http://pages.pomona.edu/~sg064747
Another Contender?
[email protected] http://pages.pomona.edu/~sg064747
Skew-Hamiltonian Matrices
Definition
If Bt = −B and Dt = −D, then
T =
(A BD At
)is a skew-Hamiltonian matrix (SHM).
If T is unitarily equivalent toa SHM, then T is UESHM.
UESHM ⇒ UET
If T is skew-Hamiltonian, then T = JT tJ∗ where
J =
(0 I−I 0
).
[email protected] http://pages.pomona.edu/~sg064747
Skew-Hamiltonian Matrices
Definition
If Bt = −B and Dt = −D, then
T =
(A BD At
)is a skew-Hamiltonian matrix (SHM). If T is unitarily equivalent toa SHM, then T is UESHM.
UESHM ⇒ UET
If T is skew-Hamiltonian, then T = JT tJ∗ where
J =
(0 I−I 0
).
[email protected] http://pages.pomona.edu/~sg064747
Skew-Hamiltonian Matrices
Definition
If Bt = −B and Dt = −D, then
T =
(A BD At
)is a skew-Hamiltonian matrix (SHM). If T is unitarily equivalent toa SHM, then T is UESHM.
UESHM ⇒ UET
If T is skew-Hamiltonian, then T = JT tJ∗ where
J =
(0 I−I 0
).
[email protected] http://pages.pomona.edu/~sg064747
Reducibility of Skew-Hamiltonian Matrices
Theorem (Waterhouse, 2004)
If T is skew-Hamiltonian, then T is similar to A⊕ At for some A.
Proposition
If T is skew-Hamiltonian and 6× 6 or smaller, then T is reducible.
Pf. Sketch, 6× 6 case.
QT = TQ and Q = Q∗ lead to a 72× 36 real linear system,which is too large to consider symbolically.
Exploiting symmetries and using various reductions, oneobtains an equivalent 30× 21 real linear system.
Mathematica’s symbolic MatrixRank command shows thatthe rank is ≤ 20, whence a nonscalar Q exists.
[email protected] http://pages.pomona.edu/~sg064747
Reducibility of Skew-Hamiltonian Matrices
Theorem (Waterhouse, 2004)
If T is skew-Hamiltonian, then T is similar to A⊕ At for some A.
Proposition
If T is skew-Hamiltonian and 6× 6 or smaller, then T is reducible.
Pf. Sketch, 6× 6 case.
QT = TQ and Q = Q∗ lead to a 72× 36 real linear system,which is too large to consider symbolically.
Exploiting symmetries and using various reductions, oneobtains an equivalent 30× 21 real linear system.
Mathematica’s symbolic MatrixRank command shows thatthe rank is ≤ 20, whence a nonscalar Q exists.
[email protected] http://pages.pomona.edu/~sg064747
Reducibility of Skew-Hamiltonian Matrices
Theorem (Waterhouse, 2004)
If T is skew-Hamiltonian, then T is similar to A⊕ At for some A.
Proposition
If T is skew-Hamiltonian and 6× 6 or smaller, then T is reducible.
Pf. Sketch, 6× 6 case.
QT = TQ and Q = Q∗ lead to a 72× 36 real linear system,which is too large to consider symbolically.
Exploiting symmetries and using various reductions, oneobtains an equivalent 30× 21 real linear system.
Mathematica’s symbolic MatrixRank command shows thatthe rank is ≤ 20, whence a nonscalar Q exists.
[email protected] http://pages.pomona.edu/~sg064747
Reducibility of Skew-Hamiltonian Matrices
Theorem (Waterhouse, 2004)
If T is skew-Hamiltonian, then T is similar to A⊕ At for some A.
Proposition
If T is skew-Hamiltonian and 6× 6 or smaller, then T is reducible.
Pf. Sketch, 6× 6 case.
QT = TQ and Q = Q∗ lead to a 72× 36 real linear system,which is too large to consider symbolically.
Exploiting symmetries and using various reductions, oneobtains an equivalent 30× 21 real linear system.
Mathematica’s symbolic MatrixRank command shows thatthe rank is ≤ 20, whence a nonscalar Q exists.
[email protected] http://pages.pomona.edu/~sg064747
Reducibility of Skew-Hamiltonian Matrices
Theorem (Waterhouse, 2004)
If T is skew-Hamiltonian, then T is similar to A⊕ At for some A.
Proposition
If T is skew-Hamiltonian and 6× 6 or smaller, then T is reducible.
Pf. Sketch, 6× 6 case.
QT = TQ and Q = Q∗ lead to a 72× 36 real linear system,which is too large to consider symbolically.
Exploiting symmetries and using various reductions, oneobtains an equivalent 30× 21 real linear system.
Mathematica’s symbolic MatrixRank command shows thatthe rank is ≤ 20, whence a nonscalar Q exists.
[email protected] http://pages.pomona.edu/~sg064747
Irreducible skew-Hamiltonians exist if n ≥ 8
Numerical Evidence
Computation indicates that generic 8× 8 SHMs are irreducible.Finding a scalable family of provably irreducible SHMs is difficult.
Proposition
For d ≥ 4, the matrix T =
(A B0 A
), where
A =
1
2. . .
d
, B =
0 1 1 · · · 1 1
−1 0 1. . . 1 1
−1 −1 0. . . 1 1
.... . .
. . .. . .
. . ....
−1 −1 −1. . . 0 1
−1 −1 −1 · · · −1 0
,
is skew-Hamiltonian and irreducible.
[email protected] http://pages.pomona.edu/~sg064747
Irreducible skew-Hamiltonians exist if n ≥ 8
Numerical Evidence
Computation indicates that generic 8× 8 SHMs are irreducible.Finding a scalable family of provably irreducible SHMs is difficult.
Proposition
For d ≥ 4, the matrix T =
(A B0 A
), where
A =
1
2. . .
d
, B =
0 1 1 · · · 1 1
−1 0 1. . . 1 1
−1 −1 0. . . 1 1
.... . .
. . .. . .
. . ....
−1 −1 −1. . . 0 1
−1 −1 −1 · · · −1 0
,
is skew-Hamiltonian and irreducible.
[email protected] http://pages.pomona.edu/~sg064747
An Instructive Example
A Potential Pitfall
T =
0 1 00 0 20 0 0
0 0 01 0 00 2 0
is UET, UECSM, and UESHM.
However, A =(
0 1 00 0 20 0 0
)itself has
none of these properties.
Remarks
If T = A⊕ At , then T is UECSM, UESHM, and UET.
Neither class UECSM nor UESHM is closed under restrictionto direct summands.
This issue is important only in dimensions ≥ 6
[email protected] http://pages.pomona.edu/~sg064747
An Instructive Example
A Potential Pitfall
T =
0 1 00 0 20 0 0
0 0 01 0 00 2 0
is UET, UECSM, and UESHM. However, A =
(0 1 00 0 20 0 0
)itself has
none of these properties.
Remarks
If T = A⊕ At , then T is UECSM, UESHM, and UET.
Neither class UECSM nor UESHM is closed under restrictionto direct summands.
This issue is important only in dimensions ≥ 6
[email protected] http://pages.pomona.edu/~sg064747
An Instructive Example
A Potential Pitfall
T =
0 1 00 0 20 0 0
0 0 01 0 00 2 0
is UET, UECSM, and UESHM. However, A =
(0 1 00 0 20 0 0
)itself has
none of these properties.
Remarks
If T = A⊕ At , then T is UECSM, UESHM, and UET.
Neither class UECSM nor UESHM is closed under restrictionto direct summands.
This issue is important only in dimensions ≥ 6
[email protected] http://pages.pomona.edu/~sg064747
An Instructive Example
A Potential Pitfall
T =
0 1 00 0 20 0 0
0 0 01 0 00 2 0
is UET, UECSM, and UESHM. However, A =
(0 1 00 0 20 0 0
)itself has
none of these properties.
Remarks
If T = A⊕ At , then T is UECSM, UESHM, and UET.
Neither class UECSM nor UESHM is closed under restrictionto direct summands.
This issue is important only in dimensions ≥ 6
[email protected] http://pages.pomona.edu/~sg064747
An Instructive Example
A Potential Pitfall
T =
0 1 00 0 20 0 0
0 0 01 0 00 2 0
is UET, UECSM, and UESHM. However, A =
(0 1 00 0 20 0 0
)itself has
none of these properties.
Remarks
If T = A⊕ At , then T is UECSM, UESHM, and UET.
Neither class UECSM nor UESHM is closed under restrictionto direct summands.
This issue is important only in dimensions ≥ 6
[email protected] http://pages.pomona.edu/~sg064747
Final Answer
Theorem (SRG, JET)
T is UET if and only if T is unitarily equivalent to a direct sum of:
1 irreducible CSMs
2 irreducible SHMs (automatically 8× 8 or larger),
3 matrices of the form A⊕ At where A is irreducible and neitherUECSM nor UESHM (automatically 6× 6 or larger).
Moreover, the unitary orbits of these classes are pairwise disjoint.
Corollary
If T ∈ Mn(C) is UET and n ≤ 7, then T is UECSM. If n ≤ 5, thenT is unitarily equivalent to a direct sum of irreducible CSMs.
[email protected] http://pages.pomona.edu/~sg064747
Final Answer
Theorem (SRG, JET)
T is UET if and only if T is unitarily equivalent to a direct sum of:
1 irreducible CSMs
2 irreducible SHMs (automatically 8× 8 or larger),
3 matrices of the form A⊕ At where A is irreducible and neitherUECSM nor UESHM (automatically 6× 6 or larger).
Moreover, the unitary orbits of these classes are pairwise disjoint.
Corollary
If T ∈ Mn(C) is UET and n ≤ 7, then T is UECSM. If n ≤ 5, thenT is unitarily equivalent to a direct sum of irreducible CSMs.
[email protected] http://pages.pomona.edu/~sg064747
Final Answer
Theorem (SRG, JET)
T is UET if and only if T is unitarily equivalent to a direct sum of:
1 irreducible CSMs
2 irreducible SHMs (automatically 8× 8 or larger),
3 matrices of the form A⊕ At where A is irreducible and neitherUECSM nor UESHM (automatically 6× 6 or larger).
Moreover, the unitary orbits of these classes are pairwise disjoint.
Corollary
If T ∈ Mn(C) is UET and n ≤ 7, then T is UECSM. If n ≤ 5, thenT is unitarily equivalent to a direct sum of irreducible CSMs.
[email protected] http://pages.pomona.edu/~sg064747
Final Answer
Theorem (SRG, JET)
T is UET if and only if T is unitarily equivalent to a direct sum of:
1 irreducible CSMs
2 irreducible SHMs (automatically 8× 8 or larger),
3 matrices of the form A⊕ At where A is irreducible and neitherUECSM nor UESHM (automatically 6× 6 or larger).
Moreover, the unitary orbits of these classes are pairwise disjoint.
Corollary
If T ∈ Mn(C) is UET and n ≤ 7, then T is UECSM. If n ≤ 5, thenT is unitarily equivalent to a direct sum of irreducible CSMs.
[email protected] http://pages.pomona.edu/~sg064747
Final Answer
Theorem (SRG, JET)
T is UET if and only if T is unitarily equivalent to a direct sum of:
1 irreducible CSMs
2 irreducible SHMs (automatically 8× 8 or larger),
3 matrices of the form A⊕ At where A is irreducible and neitherUECSM nor UESHM (automatically 6× 6 or larger).
Moreover, the unitary orbits of these classes are pairwise disjoint.
Corollary
If T ∈ Mn(C) is UET and n ≤ 7, then T is UECSM. If n ≤ 5, thenT is unitarily equivalent to a direct sum of irreducible CSMs.
[email protected] http://pages.pomona.edu/~sg064747
Final Answer
Theorem (SRG, JET)
T is UET if and only if T is unitarily equivalent to a direct sum of:
1 irreducible CSMs
2 irreducible SHMs (automatically 8× 8 or larger),
3 matrices of the form A⊕ At where A is irreducible and neitherUECSM nor UESHM (automatically 6× 6 or larger).
Moreover, the unitary orbits of these classes are pairwise disjoint.
Corollary
If T ∈ Mn(C) is UET and n ≤ 7, then T is UECSM. If n ≤ 5, thenT is unitarily equivalent to a direct sum of irreducible CSMs.
[email protected] http://pages.pomona.edu/~sg064747
Final Answer
Theorem (SRG, JET)
T is UET if and only if T is unitarily equivalent to a direct sum of:
1 irreducible CSMs
2 irreducible SHMs (automatically 8× 8 or larger),
3 matrices of the form A⊕ At where A is irreducible and neitherUECSM nor UESHM (automatically 6× 6 or larger).
Moreover, the unitary orbits of these classes are pairwise disjoint.
Corollary
If T ∈ Mn(C) is UET and n ≤ 7, then T is UECSM.
If n ≤ 5, thenT is unitarily equivalent to a direct sum of irreducible CSMs.
[email protected] http://pages.pomona.edu/~sg064747
Final Answer
Theorem (SRG, JET)
T is UET if and only if T is unitarily equivalent to a direct sum of:
1 irreducible CSMs
2 irreducible SHMs (automatically 8× 8 or larger),
3 matrices of the form A⊕ At where A is irreducible and neitherUECSM nor UESHM (automatically 6× 6 or larger).
Moreover, the unitary orbits of these classes are pairwise disjoint.
Corollary
If T ∈ Mn(C) is UET and n ≤ 7, then T is UECSM. If n ≤ 5, thenT is unitarily equivalent to a direct sum of irreducible CSMs.
[email protected] http://pages.pomona.edu/~sg064747
A Trace Criterion
Theorem (Pearcy, 1962 + Sibirskiı, 1968)
For 3× 3 matrices, A ∼= B if and only if the traces of
X , X 2, X 3, X ∗X , X ∗X 2, X ∗2X 2, X ∗X 2X ∗2X
agree for X = A and X = B.
Proposition (SRG, JET)
If T is 3× 3, then T is UECSM if and only if
tr[T ∗T (T ∗T − TT ∗)TT ∗] = 0.
Proof.
Check if T ∼= T t . Only the final word is not satisfied trivially.
[email protected] http://pages.pomona.edu/~sg064747
A Trace Criterion
Theorem (Pearcy, 1962 + Sibirskiı, 1968)
For 3× 3 matrices, A ∼= B if and only if the traces of
X , X 2, X 3, X ∗X , X ∗X 2, X ∗2X 2, X ∗X 2X ∗2X
agree for X = A and X = B.
Proposition (SRG, JET)
If T is 3× 3, then T is UECSM if and only if
tr[T ∗T (T ∗T − TT ∗)TT ∗] = 0.
Proof.
Check if T ∼= T t . Only the final word is not satisfied trivially.
[email protected] http://pages.pomona.edu/~sg064747
A Trace Criterion
Theorem (Pearcy, 1962 + Sibirskiı, 1968)
For 3× 3 matrices, A ∼= B if and only if the traces of
X , X 2, X 3, X ∗X , X ∗X 2, X ∗2X 2, X ∗X 2X ∗2X
agree for X = A and X = B.
Proposition (SRG, JET)
If T is 3× 3, then T is UECSM if and only if
tr[T ∗T (T ∗T − TT ∗)TT ∗] = 0.
Proof.
Check if T ∼= T t . Only the final word is not satisfied trivially.
[email protected] http://pages.pomona.edu/~sg064747
Bibliography
1 Balayan, L., Garcia, S.R., Unitary equivalence to a complex symmetric matrix:geometric criteria, Oper. Matrices 4 (2010), No. 1, 53–76.
2 Garcia, S.R., Poore, D.E., Wyse, M.K., Unitary equivalent to a complex symmetricmatrix: a modulus criterion, Oper. Matrices (to appear).
3 Garcia, S.R., Tener, J.E., Unitary equivalence of a matrix to its transpose, J. OperatorTheory (to appear).
4 Garcia, S.R., Wogen, W., Complex symmetric partial isometries, J. Funct. Analysis257 (2009), 1251-1260.
5 Garcia, S.R., Wogen, W., Some new classes of complex symmetric operators, Trans.Amer. Math. Soc. 362 (2010), 6065-6077.
6 Halmos, P.R., Linear algebra problem book, The Dolciani Mathematical Expositions, 16.Mathematical Association of America, Washington, DC, 1995.
7 Pearcy, C., A complete set of unitary invariants for 3× 3 complex matrices, Trans. Amer.Math. Soc. 104 (1962), 425429.
8 Sibirskiı, K.S., A minimal polynomial basis of unitary invariants of a square matrix oforder three (Russian), Mat. Zametki 3 (1968), 291295.
9 Tener, J.E., Unitary equivalence to a complex symmetric matrix: an algorithm, J. Math.Anal. Appl. 341 (2008), 640–648.
10 Waterhouse, William C., The structure of alternating-Hamiltonian matrices, LinearAlgebra Appl. 396 (2005), 385390.
[email protected] http://pages.pomona.edu/~sg064747